OBJECTIVE: Positron emission tomography (PET), the reference technique for in vivo noninvasive assessment of myocardial perfusion and metabolism, is hampered by limited resolution and low signal-to-noise ratio. Cardiac MR imaging, on the other hand, provides excellent soft-tissue contrast. This study examines the feasibility of combining the information of these two complementary techniques by the three-dimensional superimposition of regional myocardial blood flow or substrate metabolism as depicted in cardiac PET images on comparable MR images at the same cardiac phase and spatial location. SUBJECTS AND METHODS: Three-dimensional, gated PET and MR images of the heart were acquired at different phases of the cardiac cycle from six normal volunteers and from one patient with coronary artery disease that had been detected by coronary angiography. An interactive algorithm using morphologic operators was developed to contour the left ventricle on the MR and PET images. A three-dimensional surface-fitting technique was used to register the left ventricle surfaces. The accuracy of registration was estimated using 80 internal landmarks from six volunteer scans. RESULTS: These techniques yielded PET images resliced along the same spatial location and orientation as the MR images both in the transaxial and short-axis views. The average residual, a measure of the goodness of fit, was 26 (+/- 5.6) for the systolic and 13 (+/- 6.1) for the diastolic images compared with an increase of that index from 9.3 at the best fit to 13.2 when the images were deliberately misaligned by 2 mm in each of two directions. We verified that MR and PET images could be aligned with an accuracy of 1.95 mm (+/- 1.6), which was approximately equal to the larger of the two pixel sizes (i.e., 1.6 mm on PET images). CONCLUSION: MR and PET images of the heart at identical cardiac phases can be accurately superimposed. Both transaxial and short-axis views can be obtained, the latter being more useful for PET quantification. This technique offers the potential for characterizing regional interactions among contractile function, blood flow, and substrate metabolism, especially when these are altered regionally in cardiac diseases.
OBJECTIVE: Positron emission tomography (PET), the reference technique for in vivo noninvasive assessment of myocardial perfusion and metabolism, is hampered by limited resolution and low signal-to-noise ratio. Cardiac MR imaging, on the other hand, provides excellent soft-tissue contrast. This study examines the feasibility of combining the information of these two complementary techniques by the three-dimensional superimposition of regional myocardial blood flow or substrate metabolism as depicted in cardiac PET images on comparable MR images at the same cardiac phase and spatial location. SUBJECTS AND METHODS: Three-dimensional, gated PET and MR images of the heart were acquired at different phases of the cardiac cycle from six normal volunteers and from one patient with coronary artery disease that had been detected by coronary angiography. An interactive algorithm using morphologic operators was developed to contour the left ventricle on the MR and PET images. A three-dimensional surface-fitting technique was used to register the left ventricle surfaces. The accuracy of registration was estimated using 80 internal landmarks from six volunteer scans. RESULTS: These techniques yielded PET images resliced along the same spatial location and orientation as the MR images both in the transaxial and short-axis views. The average residual, a measure of the goodness of fit, was 26 (+/- 5.6) for the systolic and 13 (+/- 6.1) for the diastolic images compared with an increase of that index from 9.3 at the best fit to 13.2 when the images were deliberately misaligned by 2 mm in each of two directions. We verified that MR and PET images could be aligned with an accuracy of 1.95 mm (+/- 1.6), which was approximately equal to the larger of the two pixel sizes (i.e., 1.6 mm on PET images). CONCLUSION: MR and PET images of the heart at identical cardiac phases can be accurately superimposed. Both transaxial and short-axis views can be obtained, the latter being more useful for PET quantification. This technique offers the potential for characterizing regional interactions among contractile function, blood flow, and substrate metabolism, especially when these are altered regionally in cardiac diseases.
Authors: Riemer H J A Slart; Jeroen J Bax; Dirk J van Veldhuisen; Ernst E van der Wall; Rudi A Dierckx; Jaep de Boer; Pieter L Jager Journal: J Nucl Cardiol Date: 2006 Mar-Apr Impact factor: 5.952